Detection of adamantane-sensitive influenza A(H3N2) viruses in Australia, 2017: a cause for hope?

For over a decade virtually all A(H3N2) influenza viruses have been resistant to the adamantane class of antivirals. However, during the 2017 influenza season in Australia, 15/461 (3.3%) adamantane-sensitive A(H3N2) viruses encoding serine at residue 31 of the M2 protein were detected, more than the total number identified globally during the last 6 years. A return to wide circulation of adamantane-sensitive A(H3N2) viruses would revive the option of using these drugs for treatment and prophylaxis.

For over a decade virtually all A(H3N2) influenza viruses have been resistant to the adamantane class of antivirals. However, during the 2017 influenza season in Australia, 15/461 (3.3%) adamantane-sensitive A(H3N2) viruses encoding serine at residue 31 of the M2 protein were detected, more than the total number identified globally during the last 6 years. A return to wide circulation of adamantane-sensitive A(H3N2) viruses would revive the option of using these drugs for treatment and prophylaxis.
Amantadine and rimantadine are compounds of the adamantane class of antivirals which act on influenza A viruses by binding to the M2 ion channel, preventing uncoating of the virus during replication. Treatment of influenza A virus infection with these drugs within 48 hours of symptom onset reduces illness by ca 24 hours, and when given prophylactically, the drugs can prevent ca 60% of influenza cases [1]. However, effectiveness of both drugs is lost when viruses acquire an amino acid substitution at one of five critical residues of the M2 protein i.e. positions 26, 27, 30, 31 and 34 [2]. The occurrence of adamantane-resistant influenza A viruses was rare among circulating influenza viruses [3] until 2000, when an increasing proportion of viruses from Asia, particularly China, contained the serine (S) to asparagine (N) substitution at residue 31 (S31N) of the M2 protein [4]. By the end of the 2005/06 influenza season, over 90% of circulating A(H3N2) viruses in North America and other parts of the northern hemisphere, such as Asia, contained the S31N substitution even though the vast majority of resistant viruses were from patients who had not been treated with adamantanes [5]. After more than 7 years of almost complete dominance of adamantane-resistant A(H3N2) influenza viruses globally, we describe the detection in Australia of increased numbers of adamantane-sensitive viruses during the 2017 influenza season.

Global frequency of adamantane-resistant A(H3N2) viruses 1968-2017
Analysis of all M2 gene sequences of A(H3N2) viruses deposited in the EpiFlu public sequence database of the Global Initiative on Sharing All Influenza Data (GISAID) (n = 26,231; as at 21 November 2017), showed that the global frequency of adamantane-resistant A(H3N2) viruses with M2 N31 has been > 99% each year from 2010 to 2017 (Table 1).

Analysis of the adamantane-resistance situation in Australia during the 2017 influenza season
The Australian 2017 influenza season was dominated by high levels of A(H3N2) influenza virus activity. During this season, 15 adamantane-sensitive A(H3N2) viruses encoding M2 S31 were detected in Australia (Table 2), which exceeded the cumulative total of 11 adamantane-sensitive influenza A viruses detected globally between 2011 and 2016 (Table 1). In contrast, the frequency of adamantane-resistance in circulating A(H1N1)pdm09 viruses has remained unchanged at > 99.9% both in Australia and worldwide.
In Victoria, M2 S31 viruses were detected in July, August and September, with the peak detection frequency of five of 35 occurring in August ( Figure 1). In addition to the 10 M2 S31 viruses, a further five Victorian A(H3N2) viruses contained an alternative amino acid, aspartic acid (D), at residue 31 of the M2 protein (M2 D31 viruses) ( Figure 1)( Table 2). M2 D31 viruses have been extremely rare since the emergence of the A(H3N2) viruses in 1968, with just 11 of 21,064 (0.05%) viruses detected globally before 2017, compared with the detection rate of five of 210 (2.4%) seen between July and September 2017 in Victoria ( Figure 1). However, M2 D31 viruses remain adamantane-resistant [9]. Original specimens were available for eight of the M2 S31 or D31 virus isolates, and sequencing confirmed 100% nt match between the isolates/clinical specimen pairs, with no evidence of mixed viral populations at the codon for position 31 of the M2 protein.

Phylogenetic analysis of Australian M2 S31 influenza A(H3N2) viruses
To understand whether the M2 S31 or D31 viruses emerged from a single source or occurred sporadically, phylogenetic trees were constructed using the haemagglutinin (HA), neuraminidase (NA) and M2 gene sequences of the 2017 A(H3N2) viruses from Australia, together with global A(H3N2) virus sequences from March to September 2017 available from GISAID ( Figure  2, Figure 3 and Figure 4). The authors acknowledge the originating and submitting laboratories that provided

Figure 1
Frequency of M2 S31 and D31 viruses among influenza A(H3N2) viruses, Australia and New Zealand, July-September 2017

Key
No M2 S31 or M2 D31 viruses were detected in the states of South Australia, Western Australia, Tasmania or the Australian Capital Territory where the number of influenza A (H3N2) viruses tested was 149, 40, 52 and 32, respectively.  Australian M2 S31 viruses are coloured in blue, non-Australian M2 S31 viruses are coloured in green, and Australian M2 D31 viruses are coloured in red. All other viruses contain the N31 in the M2 protein. GISAID ID numbers for the M2 S31 and D31 virus sequences are listed in Table 2. The sequence data were aligned using the MAFT algorithm in Geneious V10.0.09 (Biomatters Ltd, Auckland, New Zealand). Maximum likelihood phylogenetic trees were generated with RAxML v8.2, using nucleotide substitution models. Amino acid substitutions were calculated using BASEML within PAML 4.9e with FigTree v1.4.3 used to project amino acid substitutions onto the phylogenetic trees.  Table 2. The sequence data were aligned using the MAFT algorithm in Geneious V10.0.09 (Biomatters Ltd, Auckland, New Zealand). Maximum likelihood phylogenetic trees were generated with RAxML v8.2, using nucleotide substitution models. Amino acid substitutions were calculated using BASEML within PAML 4.9e with FigTree v1.4.3 used to project amino acid substitutions onto the phylogenetic trees. sequences to GISAID's EpiFlu database which were used to construct the phylogenetic trees (www.gisaid. org). In both HA and NA phylogenetic trees all but two of the Australian M2 S31 viruses formed a monophyletic group, together with a Japanese and New Zealand M2 S31 virus. The Japanese virus, A/Mie/18/2017 (isolated in May 2017), was positioned ancestrally to the M2 S31 Australian viruses in both the HA and NA clades (Figure 2 and Figure 3), suggesting it may have been the progenitor. All of the M2 S31 viruses were part of the HA clade 3C.2a1, except for one strain (A/ Victoria/22/2017) from July which was a 3C.2a virus (Figure 2), and all but one (A/Victoria/211/2017) of the 3C.2a1 viruses formed the subclade characterised by HA amino acid substitutions R261Q and K83R ( Figure  2), which exclusively contained these viruses even when all A(H3N2) viruses via GISAID were included in the analysis. The M2 D31 viruses clustered together phylogenetically, forming a separate branch within the HA clade 3C.2a1, although this clade also contained some M2 N31 Australian viruses (Figure 2).

Discussion and conclusion
Adamantane-resistance was first detected in persons infected with influenza virus who were treated with adamantanes in the late 1980s, in closed settings, such as nursing homes [10,11], as well as community settings [12]. In the latter, there was apparent transmission of drug-resistant strains, when rimantadine was used for treatment or post-exposure prophylaxis in families [12]. Adamantane-resistant viruses were also detected in nursing home patients without exposure to these drugs [13], demonstrating that adamantaneresistant variants may be able to spread in the community. Up to 45% of children treated with rimantadine have been reported to shed resistant viruses [14]. In addition to widespread adamantane-resistance among A(H3N2) influenza viruses circulating globally, seasonal influenza A(H1N1) viruses also developed adamantane-resistance between 2005 and 2008 [15,16], and the A(H1N1)pdm09 virus that emerged and caused the influenza pandemic in 2009 was also adamantaneresistant. As consequence of these developments, adamantanes are no longer recommended for treatment of influenza [5].   The spread of M2 N31 viruses in the early 2000s was thought to be due to linkage to advantageous substitutions elsewhere in the virus, in a process referred to as genetic 'hitch-hiking', and not related to adamantane-induced selection pressure [17]. Even though the A(H3N2) virus has continued to undergo substantial antigenic and genetic evolution over the last decade, the M2 N31 residue has remained almost completely fixed, suggesting that during that time it contributed to viral fitness.  the manuscript. Y-MD, MK, NK, SS, KS, IB and AH contributed to the design and interpretation of the results, reviewed the early draft and approved the final version.